To pilot or not to pilot?
One of the important definitions before an implementation project is if a piloting test should be performed or not. Some of the considerations for this decision are:
– A pilot test will provide a certain way to double check the results of the different processes within the company, from the legacy system to the new system
– A pilot test will in fact double the workload for end users, as the tasks will have to be performed once in the legacy system and another time in the new system
– The verification process of the results can be a very time consuming process
– If there are substantial changes to be done on the structure of the data (such as changes on chart of accounts or structure of key elements such as customer, item and vendor keys, for example), there will be an added layer of complexity in order to define cross reference data for the verification process
It has been our experience that there is a different approach that can be as productive as a pilot testing but that can reduce the amount of effort to be performed. We usually request for our users to define a list of scenarios that are produced in their daily operations (especially the crucial, complex type of scenarios), and to create a documentation set for each one of them.
Once this is defined, we use our sandbox database to create all required elements to process such scenarios, one by one. The documentation provided should contain reports that indicate what the results of the specific scenario should be, and this can be used to compare the output from the new system. These documents are also a good basis for sign off on different steps of the project.